NEW MULTIVARIATE TECHNIQUES FOR CALCULATION OF BLOOD ANALYTE CONCENTRATIONS

用于计算血液分析物浓度的新多元技术

• 批准号: 6314186
• 负责人: ANDREW BERGER
• 金额: $ 1.38万
• 依托单位: MASSACHUSETTS INSTITUTE OF TECHNOLOGY
• 依托单位国家: 美国
• 财政年份: 2000
• 资助国家: 美国
• 起止时间: 2000-06-01 至 2001-05-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/6314186
• 关键词: bioengineering /biomedical engineering; biological products; biomedical equipment development; biomedical resource; diabetes mellitus; human tissue; model design /development; statistics /biometry; technology /technique

In the spectra of most biological samples, there are overlapping bands from different species. While standard multivariate calibration methods, most notably PLS, can extract individual analytes' concentrations with impressive accuracy, they cannot utilize the known spectrum of an analyte during the calibration process. We have recently developed a new calibration method called hybrid linear analysis (HLA) that incorporates the known spectrum into the calibration. Already we have seen that this method can generate equivalent or superior calibrations to PLS in predicting glucose in various media including blood serum. In our upcoming work, we will extend the application of this method to other analytes, as well as continue to explore the fundamental properties of the algorithm itself.

在大多数生物样品的光谱中，存在重叠 来自不同物种的乐队。 而标准的多元校准 方法，最值得注意的是可以提取单个分析物的方法 浓度具有令人印象深刻的精度，他们无法利用已知的 在校准过程中分析物的频谱。 我们有 最近开发了一种称为混合线性的新校准方法 分析（HLA）将已知频谱纳入 校准。 我们已经看到这种方法可以生成 在预测葡萄糖中的同等校准或优质的校准 包括血清在内的各种培养基。 在即将到来的工作中，我们将 将此方法应用于其他分析物以及 继续探索算法的基本属性 本身。